1 Introduction
When an acoustic pressure wave travels down the ear canal toward the TM, the frequency-
1
dependent power P(f ) [W],1 which is defined as the flow of energy over time and space,
2
is continuous until it reaches an impedance discontinuity, such as a change in the area of
3
the ear canal, or the tympanic membrane (TM). Such variations in the wave propagation
4
result in frequency dependent reflections. The ratio of the reflected to absorbed power is
5
called the power reflectance.The ratio of the reflected to absorbed pressure (or velocity) is
6
called the complex pressure reflectance. Historically reflectance is denoted by the upper
7
case Greek letter “gamma” Γ(f ).2
8
1.1 Absorbed and reflected waves
9
The key behind the reflectance measure is to decompose the acoustic wave into absorbed
10
and reflected components. The notation is simplified if we relabel the absorbed and
11
reflected waves in terms of their direction of travel. Thus we define pressures P+(f ) =
12
Pa(f ) and P−(f ) = Pr(f ) where the “±” indicates the direction of travel of the wave.
13
This results in the pressure and velocity relations
14
P (f ) = P+(f ) + P−(f ) (1)
15
and
16
U (f ) = U+(f ) − U−(f ). (2)
17
The pressure (like voltage) is a scaler, while the velocity (like current), is a vector, with
18
direction. This explains the change in sign of Eq. 2.
19
At each point x along the canal, the ratios of the pressures P± and velocities U± are
20
equal to the acoustic resistance, which due to conservation of energy must always be
21
positive and real. Thus
22
ρc
A(x) = P+(f )
U+(f ) = P−(f )
U−(f ), (3)
23
where ρc = 407 [Rayls] and A(x) is the canal area at location x along the ear canal.
24
If we reorganize the above equation we obtain the definition of the complex reflectance
25
Γ(f ) = P−(f )
P+(f ) = U−(f )
U+(f ), (4)
26
in terms of the ratio of reflected over absorbed pressure (or velocity). This relationship
27
is complex, thus Γ may be written either as the sum of a real and imaginary part, or in
28
terms of its magnitude and phase3 (Eq. A.4).
29
1For clarity all SI units will be displayed in square brackets.
2This article assumes the reader knows mathematics at the level of high school algebra. In keeping with this assumption, we will use mathematical notation, which frequently includes Greek letters. Using mathematical notation is simpler because it precisely specifies the variables via a unique symbol.
3Given the nature of the topic it is necessary to use complex numbers to represent pressure and velocity, and their ratio, which defines the impedance at a given frequency. For example, the algebraic sentence Γ(f ) = |Γ(f )|ejφ says that the complex reflectance Γ(f ), as a function of frequency f , is equal to the magnitude of Γ at each frequency, times the exponent of the phase φ, or angle. Here ejφ= cos(φ) + j sin(φ) is called Eulers formula (j =√
−1).
The beauty of complex reflectance over the complex impedance is that the magnitude
30
reflectance |Γ(f )| describes the relative absorbed power as a function of frequency, while
31
the phase codifies the latency of the absorbed power (e.g., depth of the reflected wave).
32
Furthermore complex reflectance has certain properties that render it more useful than
33
impedance in forming a diagnosis of a ME pathology.
34
Thus the complex reflectance has an intuitive interpretation, whereas most people
35
find complex impedance difficult to master. Within the clinical audiology community it
36
is well know that impedance is a difficult concept. Part of the problem is the obtuse
37
terminology, as outline in Table B1. But terminology is not the entire story. Even
38
those who say they understand complex impedance find it difficult to explain to lay
39
persons. One obvious reason for this is that the transformation from complex reflectance
40
to complex impedance is complicated [Voss and Allen, 1994]. Other than for the simplest
41
of cases, this transformation is mathematically challenging.4
42
It should be noted however that, regardless of the complexity of the relationship,
43
complex impedance and reflectance are mathematically equivalent. Since reflectance is
44
easily understood, and impedance is not, we shall focus on the complex reflectance in
45
this report.
46
Estimating Γ(f ) requires a special calibration of the earphone, known in the Engineer-
47
ing literature as a Th´evenen calibration. While quite important, this measure is beyond
48
the scope of the present discussion.5
49
Voss and Allen [1994] were the first to use the complex reflectance to characterize the
50
acoustic properties of the TM, by approximately removing the effect of the residual ear
51
canal. When the area of the canal is constant the TM reflectance Γtm(f ) is related to
52
the measured microphone reflectance Γm(f ) by the relation
53
Γm(f ) = Γtme−j2πf 2L/c (5)
54
where L is the distance from microphone to TM and c is the speed of sound. Simply
55
stated, this equation says that there is a round trip delay of 2L/c between the measure-
56
ment point and the TM. Thus if one can estimate L they then can compute Γtm(f ), which
57
is mathematically equivalent to the TM impedance. But there is one further complica-
58
tion. When the area of the canal depends on position, the effective length L depends on
59
frequency. This complicates this relationship, for both practice and theory.
60
Thus if one could estimate L, then one may estimate Γtm(f ) from the measured
61
Γm(f ). An accurate procedure for doing this that includes the case of variable canal area
62
was reported by Robinson and Allen [2013]. For most clinical applications Γtm(f ) is the
63
starting point for the analysis of ME pathologies.
64
The remainder of this chapter is focused on clinical applications of complex reflectance
65
and the effect of ME pathologies.
66
1.2 Clinical applications of the complex reflectance
67
Next we discuss some of the main applications of the complex reflectance method. The
68
first is the Forward pressure level which is a method for calibrating the signal in the
69
ear canal so that the pressure delivered to the cochlea is approximately constant, thus
70
4This transformation is known as a M¨obius, or bilinear, transformation.
5http://en.wikipedia.org/wiki/Thevenin’s_theorem
100 101
−25
−20
−15
−10
−5 0
FPL correction factor: |1+Γ|/2
20 log 10(|1+Γ|/2) [dB]
f [kHz]
Figure 1: Forward pressure level normalization factor which corrections the ear canal standing wave. At the null frequency f0, the phase of the complex reflectance (i.e., the angle of Γ(f0)) is 180 degrees [Withnell, 2010, 2009b]. This explains the source of the standing wave nulls, and their unpredictable nature, since the frequency critically depend on the round trip delay from the source back to the microphone (Eq. 5). This delay is different for each ear, as it dependent on the insertion depth of the probe and the geometry of the canal and TM.
removing troubling ear canal standing waves. A second application is a method to remove
71
the effect of the residual ear canal, defined as the section of canal between the microphone
72
and the TM. Clinically what we really need to know is the reflectance at the TM. This
73
is distorted by the round trip delay of the residual ear canal. Finally, given the complex
74
reflectance at the TM, it is possible, in theory at least, to diagnose most, if not all, middle
75
ear pathologies.
76
What the research clinician would most like to know is the transmission properties
77
of the ME. Specifically is the ME clinically normal? If not, in what way is it abnormal?
78
There is a host of possible abnormalities. The most common are: 1) TM abnormalities,
79
2) Ossicular disruption and ossification, 3) Fluid in the middle ear, 4) Eustachian tube
80
dysfunction. There is building evidence that the TM complex reflectance can be used
81
to both identify, and then quantify (diagnose) most, if not all of these pathologies. We
82
shall present the evidence in support of this hypothesis.
83
Forward pressure level and its applications: Real ear measurement systems use a
84
microphone in the ear canal to account for variation in the subject’s ME. This methods is
85
intended to replace traditional but less reliable coupler calibrations, which do not account
86
for these natural variations. In principle this seems like a logical thing to do. However
87
there are several serious problems that need to be addressed. The most obvious of these
88
is the effect of standing waves in the canal [Siegel, 1994]. While the canal standing wave
89
problem has been recognized for a long time, there have been years of debate as to how
90
to best deal with it. There was considerable uncertainty in the beginning, as to the best
91
quantity to normalize, with the intensity being one obvious candidate (Gorga and Neely,
92
1998; Souza, 2014; Keefe et al.).
93
The nature of these standing waves is shown in Fig. 1. Based on the deep nulls
94
seen in the figure, clearly would not be reasonable to normalize the microphone pressure
95
Pm(f ) to be a constant. Such a normalization would boost the level at the standing
96
wave null by as much as 25 [dB]. The frequency of this boost would depend on the
97
microphone placement. However this is precisely what is done in real ear measurements
98
when the microphone level is re-normalized to be constant. Real-ear calibrations have
99
the unintended effect of putting a large peak at the standing wave frequency.
100
But it now appears that this question has been satisfactorily resolved by the research
101
community. It is now recognized that the forward pressure level (FPL) should be made
102
constant [Souza, Dhar, Neely, Siegel 2014 (9 methods; JASA)]. However, other than some
103
research systems, FPL has not made it into clinical equipment at the time of this writing.
104
This calibration procedure is a variant on the so-called RETSPL method, where the
105
voltage on the transducer is varied to maintain the pressure at the TM constant in an ar-
106
tificial ear. Here we coin the term RETFPL (reference equivalent forward pressure level).
107
The procedure is: Given the microphone pressure Pm(f ) and its corresponding complex
108
reflectance Γm(f ), compute the forward pressure P+(f ). Vary the receiver voltage so
109
that P+(f ) is constant at the desired level.
110
One may determine P+(f ) as follows:
Pm(f ) = P+(f ) + P−(f )
= P+(f )
1 + P−
P+
= P+(f )(1 + Γm(f )).
Solving for P+ gives
111
P+(f ) = Pm(f )
1 + Γm(f ). (6)
112
Since the microphone pressure and the complex reflectance are both known, one may
113
easily solve for the absorbed (i.e., forward) pressure by the above relation. While some
114
alternative methods have been tried, it seems unlikely that, without knowledge of the
115
complex reflectance Γ(f ), the precision needed to estimate the standing wave frequency
116
will be inadequate.
117
Fig. 6 shows the normalization factor for 10 human ears, taken from [Voss and Allen,
118
1994]. As the frequency is increased above 3 [kHz], the phase plays a very important
119
role, as it can result in ear canal standing waves, as seen by the canal microphone. This
120
phase relation is negligible at low frequencies, where it is small. At higher frequencies,
121
the phase, due to the round trip delay from Tagus to TM (2.5 cm) of ≈150 [µs], gives a
122
standing wave null at ≈3.4 [kHz]. If one includes the additional delay in the TM (≈45
123
µs) and ossicles, the frequency null can be even lower. Examples of these nulls, as shown
124
in Fig. 1, have been computed from the complex reflectance using the relation
125
FPL = 20 log10
1 + Γ(f ) 2
.
126
The extra factor of 2 is to compensate for the fact that Γ(f ) goes to 1 as the frequency
127
goes to zero. Thus the sum of the two terms is 2, which is removed by this factor. The
128
use of this extra factor is optional depending on the interpretation of what is desired at
129
low frequencies. This factor is know in the literature as the missing 6 dB.
verify this claim.
130
To experimentally verify this formula consider the following experiment. If we mea-
131
sure the pressure at the end of a rigid cavity, and normalize P+(f ) to be a constant
132
(say 1 [Pa]), then at the end of the cavity, Γ(f ) = 1 (since the cavity is rigid). Thus
133
P−(f ) = P+(f ) = 1. It follows that the pressure at the end of the cavity is 2 since
134
P = P++ P−= 1 + 1 = 2 [Pa].
135
In summary: When a earphone excites sound in the ear canal, a forward traveling
136
wave is launched into the canal. This wave proceeds down the canal at the speed of
137
sound. Once it reaches the TM (or any other change in impedance) it is reflected by an
138
amount given by the TM reflection coefficient Γtm(f ), in a frequency-dependent manner.
139
This results in a backward propagating wave P−(f ) = Γtm(f )P+(f ). All of the acoustic
140
physics of the TM is captured by Γm(f ). When the microphone measurement location
141
is not at the TM, there is a phase factor given by the round-trip delay between the TM
142
and the microphone location which is the source of the standing wave.
143
TM Acoustic admittance: From the clinical diagnostic point of view, the most im-
144
portant quantity is the acoustic admittance of the TM Ytm(f ). This is what tympanome-
145
try does not, and cannot accurately estimate. While the low frequency canal admittance
146
magnitude may be measured (this is what 226 [Hz] tympanometry provides), Ytm(f ) is
147
not measured at relevant (i.e., speech) frequencies. Furthermore, the complex TM admit-
148
tance is not accurately estimated from the canal admittance, even at 226 [Hz], because
149
the clinical procedure does not accurately remove the effect of the residual ear canal
150
[Shanks et al, 1981; Rabowitz, 1981; Shanks et al., 1988].
151
There are several challenges here. Clinically speaking, we would like to estimate the
152
wideband complex admittance at the TM given pressure measurements at some unknown
153
location in the canal. Note we do not know either 1) the length of the ear canal, nor 2)
154
the pressure at the TM. One might wonder if this is even a realistic goal.
155
What research clinicians would most like to know is the wideband complex reflectance
156
at the TM, Γtm(f ) = P−(f )/P+(f ). As previously stated, given Γtm(f ) the complex
157
TM admittance or impedance are easily determined. The proposal then is to Thevenin
158
calibrate the ear canal probe. Then given the ear canal pressure Pm(f ) one may compute
159
the complex reflectance Γm(f ). Using the procedure of Robinson and Allen (2013), the
160
phase factor corresponding to the residual ear canal may be uniquely determined and
161
removed, leaving the desired Γtm(f ). This then puts us at the starting point for the ME
162
diagnosis, as reviewed next.
163
1.3 Overview of Middle Ear Impedance and Reflectance
164
The middle ear is remarkable for its functionality in converting air-born sound into fluid-
165
born sound, via the mechanical action of the TM and ossicles. Air-borne sound that
166
reaches the TM is converted to acoustic vibration of the TM. This vibration drives the
167
ossicles that transmit the acoustic signal to the annular ligament in the oval window
168
which, in turn, convey the signal to the fluid filled cochlea.
Perhaps mention the faults analogy of a flowing river, with a dam and tributaries.
169
A middle ear model The middle ear may be represented as a simple circuit, as shown
170
in Fig. 2. At low frequencies (i.e, 200 [Hz]), the delay may be ignored, and this model
171
reduces to a parallel combination of the compliance of the ear canal, the compliance
172
of the ossicular joints and the compliance of the annular ligament. Around 0.7-1 [kHz]
173
the impedance of the cochlea, which is well approximated by a constant resistance, is
174
approximately equal to the impedance of the annular ligament, bringing the cochlear
175
load into the picture. At higher frequencies, the delays of the several transmission lines
176
may not be ignored and the full model must be considered.
177
Ear Canal Outer Ear
Ear Drum
Annular Ligament
≈37 µs delay
Mmalleus Mincus Mstapes
Pcochlea
Zcochlea
Figure 2: This shows a transmission line model of the middle ear including the outer ear (pinna and concha), the ear canal, represented as a tube, the tympanic membrane (ear drum), which is also a transmission line with approximately 37 ]µs] of delay. Following this are the ossicles consisting of the mass of the malleus, incus and stapes, shown here as electrical inductors.
Between each of these mass’s is the ligament which is represented as a spring, or in the electrical analog, a capacitor. The annular ligament holds the stapes footplate in the oval window. This spring is nonlinear since it changes its compliance when force is applied via the stapedius muscle.
Finally the cochlear impedance is represented as a resistance. This is also nonlinear in that it generates acoustic otoacoustic emissions, which are minuscule nonlinear retrograde signals (i.e., OAEs), close to the threshold of hearing for the normal ear. In this figure quote delay across and along canal. Add speaker source, for clarity, and to help raise the concept of the Thevenin source pressure and impedance, due to loading of the speaker by the ear canal. The speaker source impedance/admittance must be close to real given its very small size and internal structure [Kim].
The rapid growth in the use of otoacoustic emissions (OAEs) for hearing screening and
178
related diagnostic evaluations, has focused attention on the need for improved methods
179
of ME assessment [2]. The status of the ME is of key importance for the measurement of
180
otoacoustic emissions, since the external signal must first travel into the cochlea, followed
181
by the evoked retrograde otoacoustic emissions. Both are subject to attenuation from an
182
abnormal ME.
183
In the normal ME sounds are thus transmitted efficiently from the canal to the cochlea.
184
Relatively little acoustic power is lost in the normal middle ear (<3 [dB]) (Allen, 1986).
185
Research has shown that the middle ear is a cascade of transmission lines that are approx-
186
imately matched [Zwislocki, 1947; Moller 1965; Allen, 1986; Puria and Allen, 1991,1998].
187
This is expected, since loss would translate to poorer hearing. When this cascade of trans-
188
mission becomes unbalanced, reflections occur, and transmission is impaired. For this
189
reason the middle ear transfer of energy is very sensitive to any unbalanced impedance
190
elements within the ME structure.
191
The auditory system is extremely sensitive. The threshold of hearing at 1 [kHz] is ≈0
192
sound pressure level [dB-SPL]. This intensity corresponds to an average motion of the
193
stapes of ≈8 [pm] (i.e., ≈1/3 the radius of the Hydrogen atom (25 [pm]).
194
Due to the reciprocal nature of the middle ear, little power is lost for retrograde
195
signals; i.e., signals traveling in the reverse direction, from the oval window to the TM,
196
have similar loss characteristic as the forward traveling waves (Allen and Fahey, 1992). It
197
follows that minute vibrations generated in the cochlea by nonlinear motions of the outer
198
hair cells, can be measured in the ear canal. The measurement of these signals, known
199
as otoacoustic emissions (OAEs), are an important tool for studying the mechanism of
200
hearing.6 This has led to the development of powerful new techniques for the objective
201
assessment of hearing, such as cost-effective methods of world-wide hearing screening
202
6These signals were discovered by David Kemp and Duck on Kim, as first reported in the early 1970s.
programs.
203
The ME is also remarkably robust. Sounds of extremely high intensity (on the order
204
of 120 dB SPL) do not damage it. The inner ear, in contrast, is subject to substantial
205
damage from sounds of this intensity (the cilia of the outer hair cells are most easily
206
damaged, which results in sensorineural hearing loss). The ME provides an effective
207
protective mechanism in the form of two middle ear efferent systems. The best understood
208
is the acoustic reflex via the stapedus muscle, which helps protect the cochlea from intense
209
low-frequency vibrations. This mechanism protects the inner ear from intense air-born
210
sound, but only to a limited extent; the acoustic reflex is too slow to protect the inner ear
211
from intense sounds of short duration [Add references on combat blast levels observed,
212
which can destroy the TM, i.e., 160 dB]. A second efferent system is the tensor tympani,
213
that connects to the long process of the malleus. This system may be used during
214
speaking and chewing.
215
1.4 Impedance Discontinuities cause Reflected Acoustic Power
216
The mechanical load on the TM is from the ossicles; thus, when one says the impedance
217
of the middle ear, what one really means is the ear-canal impedance at the microphone
218
location, which is a delayed version of the drum impedance that includes the impedance
219
load of the ossicles and cochlea (Fig. 2). Power reflectance varies as a function of fre-
220
quency and depends on how the acoustic impedance of the TM varies with frequency.
221
At frequencies below 1 [kHz], the impedance of the TM is due mostly to the stiffness of
222
the annular ligament [Lynch, 1981; Allen, 1986; 13]. When pressure waves at these low
223
frequencies reach the stapes, almost all of their power is briefly stored as potential energy
224
in the stretched ligament and then reflected back to the ear canal as a retrograde pres-
225
sure wave. At even lower frequencies (<0.8 [kHz]), only a small fraction of the incident
226
power is absorbed into the ME [14]. The impedance of the TM in this frequency region
227
essentially consists of a stiffness-based reactance.
228
In a normal ear, in the mid-frequency region between 1 and 5 [kHz], the stiffness- and
229
mass-based reactances of the ME interact in a complex way and largely cancel each other.
230
The ME is a complex structure (see Fig. 2) that consists of several stiffness components
231
and several mass components that all work in harmony such that the cancellation of reac-
232
tance occurs over a wide frequency range (1 to 5 [kHz]). As a result, most of the incident
233
power that reaches the TM in this region is absorbed into the ME and transmitted to
234
the inner ear. In the low- and high-frequency regions the TM resistance is typically small
235
compared with its reactance, whereas in the mid-frequency region the resistance is larger
236
than the combination of the stiffness- and mass-based reactances.
237
The stiffness based reactance is inversely proportional to frequency (i.e., it is halved
238
with each doubling of frequency), while the resistance is constant across frequency. For
239
frequencies at the lower end of the auditory range (e.g., 0.1 [kHz]), the reactance is more
240
than 10 times the resistance, but in the region of 1 [kHz], the reactance and resistance
241
are of comparable magnitude. At frequencies above 6 [kHz], the mass-based reactance
242
of the ossicles becomes increasingly important and, because it is linearly proportional to
243
frequency, it can eventually dominate the TM impedance. This is complicated by the
244
fact that the center of mass of each ossicle barely moves. Thus the relevant mass is not
245
the ossicle mass, rather it is the rotational inertia of each ossicle.
246
To further complicate matters, our experimental knowledge of TM impedance is rel-
247
atively poor at frequencies above 6 [kHz]; thus, the frequency at which mass-based reac-
248
tance becomes the dominant component of TM impedance typically is unknown. When
249
a high-frequency pressure wave reaches the TM and mass-based reactance is substan-
250
tial, most of the power in the incident pressure wave is momentarily stored as kinetic
251
energy, primarily in the ossicles, and then reflected back to the ear canal as a retrograde
252
pressure wave. At high frequencies, much of the published data shows an impedance
253
that approaches that of a mass as the frequency is increased. Mass-based reactance is
254
linearly proportional to frequency (i.e., it doubles with each doubling of frequency), while
255
cochlear resistance varies only slightly with frequency [15]. Between 1 and 5 [kHz] the
256
resistance and reactance are of comparable magnitude, whereas at very high frequencies
257
the mass reactance can be several times greater than the cochlear resistance.
258
2 Methods
259
2.1 General properties of reflectance and admittance
260
Definitions of acoustic variables: The acoustic intensity of sound in the ear canal is
261
I(t) = p(t)u(t) [W/m2], where p(t) is the pressure in Pascals [Pa]as functions of time t.
262
The acoustic power P is the Acoustic intensity I times the cross-sectional area A [m2] of
263
the ear canal, namely P(t) = A(x) I(t) [W]. In general the area varies along the length
264
of the canal x [m]. The acoustic particle velocity [m/s] times the canal area A(x) is called
265
the volume velocity [m3/s].
266
The intensity and power may be defined either in the time or frequency domains. It
267
is important to be aware of which domain (time or frequency) is being discussed, as these
268
are very different ways of looking at I and P. Ideally we should have different notation
269
for intensity and power in the time and frequency domains. Here we shall work almost
270
exclusively in the frequency domain, where all variables shall be define in terms of a pure
271
tone of magnitude and phase [radians].
272
The ratio of the pressure over the volume velocity is the acoustic impedance, which for
273
a simple plane wave in the ear canal is ρc = 407/A(x) [Acoustic Ohms], where c = 343
274
[m/s] is the speed of sound and ρ = 1.2 [kgm/m3] is the density of air. For reference,
275
1 Pascal [Pa] is 94 [dB-SPL], which is typical of yelling. Typical speech is close to 65
276
[dB-SPL] 1 [m] from the mouth. The average threshold of hearing defines 0 [dB-SPL]
277
which is 20 [µPa], i.e., 20×10−6[Pa]. If voltage is something you feel you understand, the
278
think of the pressure as “acoustic volts.” Details of these relations are further discussed
279
in Appendix A.
280
When the ear canal area is constant, the propagation delay is constant and the phase
281
is proportional to frequency. For example, given a delay T the phase is φ = −T ω, where
282
ω = 2πf is called the radian frequency.
283
When the area depends on position (e.g., A(x)), the phase (and thus the delay)
284
depends on frequency. Fortunately this dependence, between φ(f ) and A(x), is typically
285
small. Since the phase slope defines a delay, which is related to distance by the speed of
286
sound, the phase can provide information about the location of the reflections.7 Namely
287
the frequency dependent latency can be used to determine where the reflections occur in
288
the ME. A short latency means in the middle ear or TM, while a large latency means
289
7i.e., data from Sarah R.
deep in the cochlea. Examples of this relation will be provided in later sections.
290
Furthermore the phase can tell us even more about the latency of the power absorbed.
291
Since the theory behind the complex reflectance is quite mathematical, it is not presented
292
here. However given the complex reflectance (or equivalently the complex admittance
293
or impedance), the reflections, as a function of depth, is estimated by a well defined
294
procedure.
295
Ear-canal reflectance and impedance were robustly measured with the use of the
296
multi-cavity technique first developed by Moller [1961], due to Allen [9,10,14], that makes
297
the cavity calibration method more robust to measurement problems. An ear-canal probe
298
that consists of an earphone (called a receiver in the telephone literature) and microphone
299
(aka, a transmitter) is inserted into the ear canal.
300
2.2 False positives and the middle ear
301
JLM: Remove - out of dateJBA: In what way is it out of date? Can you be more specific?
302
Lets find some clinical person that can discuss this with us, like Linda Hood? I think
303
that what they do is keep testing till they get a pass. They do try to bring the child back
304
at a later time. That does not change the statistics however. its still very expensive.
305
A major factor that contributes to the high cost of large-scale (e.g., universal) hearing
306
screening programs is the high rate of false positives. This rate is high because of the
307
inability of current screening methods to distinguish between minor conductive disorders
308
(such as a temporary blockage in the ear canal or ME) and serious inner-ear pathologies
309
(such as a sensorineural hearing loss). The practical consequences of this problem are
310
severe, since the incidence of conductive disorders is roughly 30 times greater than that
311
of inner-ear pathologies in infants [3-5].
312
Consider, for example, a universal screening program for infants: for every 1,000
313
infants screened, we expect 2 or 3 to have an inner-ear pathology (0.2%-0.3%) and 50 to
314
100 to have a conductive disorder (5%-10%). Virtually every infant with a conductive
315
disorder will be positive (fail) the screening test and subsequently require a more extensive
316
evaluation, which is expensive in both time and effort. In addition to high false-positive
317
rates, most hearing screening programs have a narrow scope. The primary objective
318
of these screening programs is to identify children with inner-ear pathologies that can
319
cause negative long-term consequences. Very few screening programs attempt to identify
320
ME pathologies in infants and newborns because of the poor reliability of instruments
321
designed for this purpose and the testing difficulty. A significant weakness of these
322
screening programs exists since chronic conductive disorders, such as otitis media with
323
effusion, also can have serious and long-term negative consequences, thus need to be
324
identified early.
325
In order for a hearing screening program to be cost-effective, the false-positive rate
326
must be substantially reduced (e.g., reduction by a factor of >30 is required to make
327
false-positive rates negligibly low and referral rates acceptable). The only possible way
328
to achieve this reduction would be with a test that can distinguish between conductive
329
disorders and inner-ear pathologies. This suggests a measure of acoustic power reflectance
330
simultaneously with otoacoustic emissions, so that we can evaluate the status of the
331
ME. Fortunately, instrumentation developed for otoacoustic emission hearing screening
332
can be modified to measure acoustic power reflectance. An instrument that combines
333
these measurements has the potential to simultaneously screen for both ME and inner-
334
ear pathologies. This would not only reduce the false-positive rate and improve cost-
335
effectiveness, it would also allow for the identification of a range of different pathologies,
336
both middle ear and cochlea.
337 JBA edits to
here
2.3 Measurement Technique
338
The earphone generates the test sound and the microphone then records either a cavity
339
or ear canal pressure. The acoustic properties of the probe system are then determined
340
with the use of frequency responses in the four rigid cavities of known impedance. The
341
ear-canal pressure and the four cavity pressures are then processed to produce an estimate Briefly
mention why 4 cavities are used.
342
of the ear-canal power and pressure reflectance.
343
Etymotic Research, Elk Grove Village, Illinois, developed the ER-10C, a distor-
344
tion product otoacoustic emission (DPOAE) ear-canal probe used in this investigation.
345
The frequency-response measurements were obtained with the use of Mimosa Acous-
346
tic’s HearID (Champaign, Illinois), an automated acoustic measurement system used for
347
DPOAE measurements. Similar methods of measuring reflectance and impedance have
348
been developed in recent years. Most of these techniques use a multi-cavity approach
349
that differ primarily in terms of the size, length and the number of calibration cavities
350
[11,16].
351
Reflectance can also be derived from measurements of acoustic impedance with other
352
methods such as the two microphone method, or the standing-wave tube method [15].
353
These various methods for measuring the acoustic impedance of the ear have been de-
354
veloped over the years [17]. Many of these early methods, which do work, have been
355
shown to be relatively inaccurate. In contrast, the four-cavity method used in this study
356
is suitable for clinical use and has been shown to be accurate from 0.1 to >8.0 [kHz], as
357
determined by measurements of standard couplers with known impedance [9] and a pair
358
of standard acoustic resistors of known resistance [10]. This improvement in accuracy is
359
mainly due to fewer assumptions. For example, the two microphone method requires a
360
precise calibration of the microphones, including their relative placement, and the stand-
361
ing wave tube method requires manual manipulation of the microphone placement. The
362
main disadvantage of the four cavity method is the need to verify the calibration each
363
day prior to any measurements, to assure that the ER-10C probe has not changed in
364
its acoustic properties. As probe technology improves, hopefully this disadvantage will
365
eventually be removed.
366
Clinical tympanometric methods have earlier been developed for measuring the re-
367
ciprocal of impedance (i.e., admittance) at a few frequencies, but the frequency range
368
of these instruments is limited to around 1 [kHz]. Recently reflectance-tympanometry
369
has been developed, mainly by Interacoustics. The utility of this composite measure is
370
presently under investigation. One of the issues with this approach is how to automati-
371
cally extract clinically useful results, and this has been the subject of on-going research
372
(Sun, Ear and Hearing articles, 2012-14). Typically tympanometry does not work in in- Better
references required 373
fant ears due to the very compliant nature of their ear canal (Margolis et al. [18]; Keefe
374
and Simmons [19]; Hunter book).
375
The reflectance measurement protocol in this study, which includes measurements of
376
acoustic impedance and related acoustic properties of the ear, uses technology that was
377
initially developed for otoacoustic emission hearing screening [20]. The measurements
378
have a bandwidth of 0.1 to 6.0 kHz, which is substantially greater than the bandwidths
379
100 101
−10
−8
−6
−4
−2 0
Absorbance: 10log 10(1−|Γ|2 ) [dB]
f [kHz]
Absorbance Test/Retest: o:S1, +:S2, x:S7, s:S9
Figure 3: Absorbance from three ears published by Voss and Allen (1994). blah blah
of most clinical instruments currently in use for assessing ME function (e.g., instruments
380
for measuring acoustic immittance). Unlike tympanometry, the system described here
381
does not vary the static pressure that is applied to the TM. While such a measurement is
382
useful for determining the static pressure behind the TM, it greatly complicates acoustic
383
power assessment.
384
Verification: This is a procedure that may be used as a check to be sure that the
385
system is working correctly. In its simplest form, one measures the reflectance of a
386
syringe or hard-walled loss-less cavity. In this case |Γ(f )| ≈ 1, which verifies the loss-less
387
condition. This is a very rigorous test, perhaps too rigorous. A deviation from 1 by a
388
few [dB] (1 dB is ≈12%) in pressure, which is not considered significant. Larger errors
389
are expected above 5 [kHz] due to small phase shifts at higher frequencies. Since an ear
390
is not a rigid cavity, perhaps a more easily interpreted test is to measure in an artificial
391
ear. However such a coupler may not be as handy as a syringe.
392
The ear-tip can also create errors. If the ear-tip is pushed against the ear canal,
393
creating a small constriction, the phase error may be introduced. Such a block can only
394
be established by replacing the ear-tip back in the canal. At times the ear-tip can become
395
clogged with crumen. In such cases, the solution is to replace the tip with a clean one.
396
Examples of the absorbance shown on a [dB] axis are given in Fig. 3.
397
2.4 Normative data sets
398
Four sets of data are described for illustration and were collected in different laboratories
399
with different power flow measurement equipment. These data are unpublished except Methods for
Normative data sets. All this needs updates.
Judi please help.
400
for the infant otitis media with effusion (OME) data. The first three data sets for the
401
adult cases (normal, otosclerosis, and perforated ear drum) were collected with the use
402
of the Reflectance Measurement System IV (Mimosa Acoustics, Inc.), which consists of a
403
digital signal processing (DSP) board, a preamplifier, an ER-10C probe, and a calibration
404
cavity set (CC4-V).
405
The OME data were collected with the Multiple Frequency Impedance and Re-
406
flectance Power Flow Analyzer system that consists of a laptop computer with a DSP
407
board, an ER-10C probe, and a similar cavity set (CC4-II). The power flow and re-
408
flectance measurements are based upon the four-cavity TM acoustic impedance measure-
409
ment methods developed by Allen [9]. The data for the three adult ears were collected
410
